AI maps Karachi’s stark green space divide, reveals urban climate vulnerability

KARACHI: A new artificial intelligence (AI) system jointly developed by researchers at New York University (NYU) and The Aga Khan University (AKU) in Karachi has exposed a stark disparity in urban green spaces across Pakistan’s largest city, highlighting critical vulnerabilities to climate change.
The study was led by Dr. Rumi Chunara, who serves as director of the NYU Center for Health Data Science and is a member of NYU Tandon’s Visualization Imaging and Data Analysis Center (VIDA), and included NYU’s Miao Zhang, Hajjra Arshad, Manzar Abbas, Hamzah Jahanzeb, Izza Tahir, Javerya Hassan and Dr. Zainab Samad from AKU. The researchers used advanced AI techniques to analyze satellite imagery and assess urban greenery in Karachi.
The research, published in the ACM Journal on Computing and Sustainable Societies, found that Karachi averages just 4.17 square meters of per capita green space, which is even less than one half of the World Health Organization’s (WHO) recommended nine square meters per person.
“It is the fifth most populous city in the world and in recent years has faced both deadly heat waves and urban flooding,” said Dr. Zainab Samad of The Aga Khan University.
“The greenspace availability varies significantly across union councils. Three union councils — Darsanno Channo, Murad Memon, and Gulshan-e-Hadeed — have the highest greenspace values of over 80 m² per capita, while five union councils — Darya Abad, Behar Colony, Chishti Nagar, Banaras Colony, and Gulshan Said — have the lowest values of less than 0.1 m² per capita.”
Dr. Samad said areas exceeding WHO’s green space recommendations were situated on Karachi’s periphery, particularly in the east.
The AI system, which achieved 89.4 percent accuracy and 90.6 percent reliability in identifying vegetation, represents a significant improvement over traditional satellite analysis, which typically achieves around 63 percent accuracy.
“To train the AI model, we create new images by shifting the hue of original satellite images,” Dr. Chunara told Arab News. “This technique helps the model better recognize diverse vegetation types.”
This “green augmentation” process enhances the model’s ability to distinguish trees from grass, even in complex urban environments, according to the expert.
The study also revealed a correlation between paved roads and increased green spaces, reflecting broader urban development patterns.
“The correlation reflects broader urban development patterns, where more developed areas with paved roads often have higher socioeconomic status, leading to better access to green spaces and urban infrastructure,” Dr. Chunara said.
The disparity in green space distribution has significant implications for public health and environmental sustainability, according to Dr. Samad. These benefits may be unequally distributed, with low-income areas often lacking vegetation that makes them hotter and more polluted.
“A combination of AI techniques could be used to not only identify deficiencies in green space but also be prescriptive about where greening could be most helpful and how this could be achieved,” she said.
The researchers emphasized the importance of making the AI system and its findings accessible to local authorities.
“Ensuring that the AI system and its findings are accessible and usable for local authorities in Pakistan is a crucial aspect of this research,” Dr. Chunara said. “We will also facilitate direct communication with local authorities to provide ongoing support and ensure that the data is effectively integrated into their planning processes.”
Policy recommendations derived from the research include prioritizing green spaces in urban planning, assessing areas where green space is most needed, and exploring potential locations for repurposing underutilized spaces for greenery.
“City planning can prioritize green spaces through master plans and zoning,” Dr. Samad said, adding that addressing the disparity in green space distribution requires interventions at various levels.
“Infrastructure initiatives like public parks and tree planting programs can enhance greenery, while community-based actions such as volunteer maintenance and tree adoption foster local involvement.”
Interestingly, the researchers compared Karachi to Singapore which, despite similar population density, provides 9.9 square meters of green space per person, exceeding the WHO target.
“In addition to Singapore, cities like Katmandu and Perth have implemented greening initiatives, such as the Green Katmandu Project and the Perth Urban Greening Strategy,” Dr. Chunara noted.
“Other cities, like Dubai, have integrated green space initiatives into their master plans to promote sustainable urban development.”
In Pakistan, however, a major challenge remains ensuring that local authorities can effectively use AI-driven research despite limited technical resources.
Dr. Chunara’s said her team was addressing this by creating accessible visualizations, data summaries and tailored reports.
“We are committed to making the findings actionable by creating clear visualizations, data summaries, and tailored reports that are easy to understand,” she said. “By engaging with authorities in a collaborative and user-friendly way, we aim to bridge the technical gap and empower them to make informed decisions for the city’s future.”